API Reference - Utilities - IterativeTrainingWrapper

Constructors

new()

Creates a new iterative training wrapper object. If there are no parameters given for that particular argument, then that argument will use default value (except for Model and CostFunction).


IterativeTrainingWrapper.new({maxNumberOfIterations: number, Model: ModelObject, CostFunctionArray: {CostFunctionObject}, targetCostValueUpperBoundArray: {number}, targetCostValueLowerBoundArray: {number}, numberOfIterationsToCheckIfConvergedArray: {number}, numberOfIterationsPerCostCalculation: number, isOutputPrinted: boolean, areUsingArraysAsInputs: boolean, iterationWaitDuration: number/boolean}): IterativeTrainingWrapperObject

Parameters:

  • maxNumberOfIterations: How many times should the model needed to be trained.

  • Model: The model to be used by the iterative training wrapper object.

  • CostFunctionArray: An array containing all the cost functions to be used by the iterative training wrapper object.

  • targetCostValueUpperBoundArray: An array containing all the upper bound of target costs.

  • targetCostValueLowerBoundArray: An array containing all the lower bound of target costs.

  • numberOfIterationsToCheckIfConvergedArray: An array containing all the number of iterations for confirming convergence.

  • numberOfIterationsPerCostCalculation: The number of iterations for each cost calculation.

  • isOutputPrinted: A boolean value that specifies if the output is printed.

  • areUsingArraysAsInputs: A boolean value that specifies if array of tensor is used.

  • iterationWaitDuration: The duration to wait between iterations. Setting it to ‘true’ will make it wait until the frame is completed.

Returns:

IterativeTrainingWrapperObject: The generated iterative training wrapper object.

Functions

setParameters() 


IterativeTrainingWrapper:setParameters({maxNumberOfIterations: number, Model: ModelObject, CostFunctionArray: {CostFunctionObject}, targetCostValueUpperBoundArray: {number}, targetCostValueLowerBoundArray: {number}, numberOfIterationsToCheckIfConvergedArray: {number}, numberOfIterationsPerCostCalculation: number, isOutputPrinted: boolean, areUsingArraysAsInputs: boolean, iterationWaitDuration: number/boolean}): IterativeTrainingWrapperObject

Parameters:

  • maxNumberOfIterations: How many times should the model needed to be trained.

  • Model: The model to be used by the iterative training wrapper object.

  • CostFunctionArray: An array containing all the cost functions to be used by the iterative training wrapper object.

  • targetCostValueUpperBoundArray: An array containing all the upper bound of target costs.

  • targetCostValueLowerBoundArray: An array containing all the lower bound of target costs.

  • numberOfIterationsToCheckIfConvergedArray: An array containing all the number of iterations for confirming convergence.

  • numberOfIterationsPerCostCalculation: The number of iterations for each cost calculation.

  • isOutputPrinted: A boolean value that specifies if the output is printed.

  • areUsingArraysAsInputs: A boolean value that specifies if array of tensor is used.

  • iterationWaitDuration: The duration to wait between iterations. Setting it to ‘true’ will make it wait until the frame is completed.

train() 


IterativeTrainingWrapper:train(featureTensor: tensor/{tensor}, labelTensor: tensor/{tensor}):

Parameters:

  • featureTensorArray: An array containing all the feature tensors.

  • labelTensorArray: An array containing all the label tensors.

Returns:

  • costMatrix: A matrix containing the cost values. The each column represents the cost from each output, while each row represents the number of iterations at which the cost values are generated.

Inherited From